The Google Cloud Next ’23 event, held in San Francisco, marked a triumphant return to in-person gatherings for the global Google Cloud community. This year’s event was an electrifying showcase of innovative product announcements, compelling customer stories, and a palpable sense of progress in the cloud computing landscape. In this comprehensive recap, we’ll delve into the pivotal moments and groundbreaking announcements from Google Cloud Next ’23.
Generative AI at the Forefront
Generative AI emerged as a focal point at Google Cloud Next ’23, particularly with the introduction of new features in Vertex AI and Duet AI. These innovations promise to open up new frontiers in cloud computing. Let’s now delve into the array of product announcements and updates unveiled during the event:
AI Platforms and Tools.
Google Cloud Next ’23 showcased a range of groundbreaking advancements in AI Platforms and Tools. Duet AI, in preview, promises to be a game-changing solution. Vertex AI Search and Conversation tools are now widely available, simplifying chatbot and custom search engine creation. Multi-Turn Search enhances user interactions with seamless follow-up questions. The Vertex AI Model Garden expands with new models like Meta’s Llama 2 and Technology Innovation Institute’s Falcon LLM, along with a preview of Anthropic’s Claude 2. PaLM 2 Foundation Model now supports 38 languages and boasts a 32,000-token context window for processing lengthy documents. Codey Chat and Code Generation Model offer up to a 25% quality improvement in code generation and chat. Imagen Image Generation Model, powered by Google DeepMind SynthID, enhances visual appeal, image editing, and captioning. Adapter Tuning in Vertex AI boosts text capabilities. Reinforcement Learning with Human Feedback (RLHF) is in public preview, enabling AI systems to learn from human input. Vertex AI Extensions enable real-time actions across applications, and Vertex AI Data Connectors streamline data ingestion. Grounding ensures generative AI outputs align with enterprise data. Google Cloud adopts the open-source unified compute framework, Ray, for scaling AI and Python workloads. Colab Enterprise, in public preview, combines Colab’s ease with enterprise-level security. Med-PaLM 2, for healthcare and life sciences, will soon be previewed. Automatic Metrics and Automatic Side by Side in Vertex AI evaluate and compare model outputs. The Vertex AI Feature Store, built on BigQuery, minimizes data duplication and maintains access policies, providing a solid foundation for AI model deployment. These advancements mark a significant stride in AI platforms and tools, set to revolutionize various industries and applications.
Google has expanded its Looker business intelligence ecosystem, emphasizing the integration of AI.
- Duet AI is now available for preview in Looker.
- The Looker semantic layer is accessible to Tableau in preview and to Microsoft Power BI in general availability.
- Looker Studio users can now directly utilize Alteryx Designer Cloud for data preparation, along with improved cloud connectivity, initially supporting Microsoft Excel and CSV files from platforms like Sharepoint and OneDrive.
- An integration with Looker partner Sisu Data, set to be available later in the quarter, aims to assist customers in pinpointing the root cause of changes, identifying data outliers, and determining the next steps for analysis.
Modern Cloud Infrastructure Supporting AI Workloads
Google’s Modern Infrastructure Cloud is tailored to swiftly, securely, and cost-effectively meet specific workload and industry requirements for businesses and governments. It encompasses a wide range of infrastructure components, from AI-optimized resources and computing to containers, networking, storage, distributed cloud, and developer tools. The infrastructure for AI workloads has seen significant advancements:
- A3 VMs, powered by NVIDIA H100 GPUs and functioning as a GPU supercomputer, will soon be generally available.
- The new Google Cloud TPU v5e, currently in preview, offers up to 2x higher training performance per dollar and up to 2.5x inference performance per dollar for LLMs and generative AI models compared to Cloud TPU v4.
- The preview of Multislice technology allows for the scaling of AI models beyond the limitations of physical TPU pods, utilizing tens of thousands of Cloud TPU v5e or TPU v4 chips.
- Support for Cloud TPUs in Google Kubernetes Engine (GKE) is now accessible for both Cloud TPU v5e and Cloud TPU v4.
- Preview support for AI inference on Cloud TPUs has also been introduced.
Collaboration Cloud: Google Workspace
Google has introduced several advancements with Duet AI across its Workspace offerings:
- Duet AI for Google Workspace is now generally available, offering a no-cost trial.
- In Google Slides, Duet AI can generate entire presentations using content from Drive and Gmail.
- In Google Meet, Duet AI provides enhancements for a polished appearance and sound quality, including studio-like features and dynamic video tiles for remote attendees.
- Duet AI in Google Meet enables real-time note-taking, action item tracking, and video snippet capture, along with features like “take notes for me,” “summary so far,” and “attend for me.”
- Smart replies in Gmail are enhanced with Duet AI for quicker and personalized responses.
- Google Chat receives a comprehensive upgrade, incorporating Duet AI. It unifies direct messages and spaces into one chronological conversation list, with features like @-mentions and starred conversations. Intelligent message prioritization based on communication patterns is slated for early next year.
- Google Chat will soon support up to 500,000 participants in a single space, facilitating large-scale community-building.
- Huddles in Chat provide a seamless integration of audio and video conversations powered by Google Meet, enhancing real-time team communication.
- More third-party and Google Workspace apps are integrated into Chat, including updated Google Drive features and support for Workday, Loom, Zoho, and LumApps. Mio will also enable message interoperability with other major platforms.
- Ahead of Next, Google announced additional AI-powered capabilities to enhance cybersecurity, implement zero trust controls, and better support digital sovereignty and compliance requirements for customers.
There is so much more! Google has pulled together a summary of all of the announcements so you can feel like you were there. Read on for the complete list of 161 announcements here on their blog. If you would like to dive deeper on the Google Cloud solutions, reach out to our Daston Google Expert team by contacting our team here or you can live chat with our team on this page.
About Daston Corporation:
Founded in 1992, Daston Corp is a Google Cloud Premier and SBA Certified HUBZone Small Business. They deliver Google Cloud Licensing and Certified Professional Support to US Public Sector Agencies, Commercial Organizations, and Educational Institutions worldwide. Daston has an award-winning team of in-house experts who can optimize Google Workspace, ensuring cost savings and adhering to best practices. They specialize in full cradle-to-grave Google Workspace deployment and migration services, serving numerous Federal, State, and Local government agencies across the US.